Fix weight decay exclusions in run_*_no‑trainer.py examples

[casinca]

What does this PR do?

fixes #42754

I'm re-using the more robust logic from trainer.py for the run_*_no_trainer.py files

https://github.com/huggingface/transformers/blob/471d7ce9abbb3bc1b3bab673367378f9dbc3caac/src/transformers/trainer.py#L1199C1-L1201C32

There are like 10 others which would require this change, some also had capital LayerNorm.weight (vs "layer_norm.weight") in no_decay = ["bias", "LayerNorm.weight"]

But before propagating to other run_*_no_trainer.py I prefer to make sure my changes are acceptable.

@wwt17 also mentioned adding to the list embeddings, I can add nn.Embedding too but this is up to 🤗 to decide.

Before submitting

• This PR fixes a typo or improves the docs (you can dismiss the other checks if that's the case).
• Did you read the contributor guideline,
  Pull Request section?
• Was this discussed/approved via a Github issue or the forum? Please add a link
  to it if that's the case.
• Did you make sure to update the documentation with your changes? Here are the
  documentation guidelines, and
  here are tips on formatting docstrings.
• Did you write any new necessary tests?

Who can review?

Anyone in the community is free to review the PR once the tests have passed. Feel free to tag
members/contributors who may be interested in your PR.

I think @zucchini-nlp and @SunMarc (for trainer) may be interested.

[casinca]

I forgot about this PR and the issue got closed for inactivity but it still seems relevant.

@Rocketknight1 👋
I think at the time you guys were busy with 5.0 and this PR was overlooked.
Could you take a look and see if it sounds good to you? or could you redirect me to the appropriate reviewer? I'm not sure my initial tagging was adequate. Thanks.

[Rocketknight1]

@bot /style

[github-actions]

Style fix bot fixed some files and pushed the changes.

[Rocketknight1]

Yes, this looks good, and sorry for the delay! The change is definitely an improvement, and I'm fine with doing something similar for the other no_trainer files if you want. We should probably just merge this PR first, though - is there anything else you want to add before I do?

[casinca]

Yes, this looks good, and sorry for the delay! The change is definitely an improvement, and I'm fine with doing something similar for the other no_trainer files if you want. We should probably just merge this PR first, though - is there anything else you want to add before I do?

Thanks for coming back to this. I'm fine as it is, I will open a separate PR for the other ones.

[HuggingFaceDocBuilderDev]

The docs for this PR live here. All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update.